The solubility and site preference of Fe3+ in Li7-3x Fe x La3Zr2O12 garnets.

A series of Fe3+-bearing Li7La3Zr2O12 (LLZO) garnets was synthesized using solid-state synthesis methods. The synthetic products were characterized compositionally using electron microprobe analysis and inductively coupled plasma optical emission spectroscopy (ICP-OES) and structurally using X-ray powder diffraction and 57Fe Mössbauer spectroscopy. A maximum of about 0.25 Fe3+ pfu could be incorporated in Li7-3x Fe x La3Zr2O12 garnet solid solutions. At Fe3+ concentrations lower than about 0.16 pfu, both tetragonal and cubic garnets were obtained in the synthesis experiments. X-ray powder diffraction analysis showed only a garnet phase for syntheses with starting materials having intended Fe3+ contents lower than 0.52 Fe3+ pfu. Back-scattered electron images made with an electron microprobe also showed no phase other than garnet for these compositions. The lattice parameter, a0, for all solid-solution garnets is similar with a value of a0≈12.98 Å regardless of the amount of Fe3+. 57Fe Mössbauer spectroscopic measurements indicate the presence of poorly- or nano-crystalline FeLaO3 in syntheses with Fe3+ contents greater than 0.16 Fe3+ pfu. The composition of different phase pure Li7-3x Fe x La3Zr2O12 garnets, as determined by electron microprobe (Fe, La, Zr) and ICP-OES (Li) measurements, give Li6.89Fe0.03La3.05Zr2.01O12, Li6.66Fe0.06La3.06Zr2.01O12, Li6.54Fe0.12La3.01Zr1.98O12, and Li6.19Fe0.19La3.02Zr2.04O12. The 57Fe Mössbauer spectrum of cubic Li6.54Fe0.12La3.01Zr1.98O12 garnet indicates that most Fe3+ occurs at the special crystallographic 24d position, which is the standard tetrahedrally coordinated site in garnet. Fe3+ in smaller amounts occurs at a general 96h site, which is only present for certain Li-oxide garnets, and in Li6.54Fe0.12La3.01Zr1.98O12 this Fe3+ has a distorted 4-fold coordination.

Introduction

Cubic Li-oxide garnet with the nominal composition Li7La3Zr2O12 (LLZO) has one of the highest Li-ion conductivities measured to date for any crystalline phase [1]. Research indicates that pure Li7La3Zr2O12 is tetragonal (space group (SG): I41/acd) at room temperature (RT) with a significantly lower ionic conductivity of about 10−6 S cm−1 than that of the high temperature cubic phase (SG: Ia-3d) that has an ionic conductivity of about 10−4–10−3 S cm−1 at RT [1]. However, in spite of all the research that has been done on LLZO, a full understanding of the stability relations between the two structures and the precise composition of “Li7La3Zr2O12” is not at hand. Three “types” of cubic LLZO have been reported: (i) a “low temperature” cubic phase which is either stabilized by CO2 or by the insertion of water molecules into the structure combined with a H+/Li+ exchange mechanism [3-5]; (ii) a high temperature cubic phase that is stable above a critical temperature, TC, of 640 °C [6]; and (iii) a cubic LLZO phase that is stable at RT by introducing small amounts of supervalent dopants such as Al3+or Ga3+ [2-10]. The respective conductivities at RT are as follows: (ii)>(iii)>(i). An undoped LLZO phase with a predicted conductivity value of 1.71×10−3 S cm−1 at RT, as obtained from conductivity values at higher temperature and then extrapolated using the Arrhenius equation [5], is the highest reported value to date.

The manner in which dopant cations are distributed over the three possible crystallographic sites containing Li in the garnet structure (Fig. 1) is thought to play a critical role in the Li-diffusion behavior. Indeed, it has been shown that the site occupation and concentration of the dopants affect the Li-ion diffusion rate in Al- and Ga-bearing LLZO garnet solid solutions [8-10].

Because aliovalent cations stabilize the high conducting cubic LLZO phase, we decided to study the solid-solution behavior of Fe3+ in garnet. Fe is a 3d transition element and it could have a different site occupation compared to Al and Ga, and thus possibly affect the Li-ion transport mechanism(s). In our previous study on Li7−3FeLa3Zr2O12 we investigated just a single composition, i.e. x=0.24. Here, we undertook a series of synthesis experiments of different Fe3+-containing Li7–3Fe3+La3Zr2O12 garnets with x=0.04–0.72. Our study mainly focuses on the solid-solution behavior, and the site preference of Fe3+ in LLZO as revealed by 57Fe Mössbauer spectroscopy.

Experimental methods

Synthesis

A series of Li7−3Fe3+La3Zr2O12 garnets with intended mole fractions (xint) of Fe3+ with xint=0.04–0.72 per formula unit (pfu) was synthesized by high-temperature sintering methods, as reported in our earlier study on Fe-bearing LLZO [11]. The starting materials were Li2CO3 (99%, Merck), La2O3 (99.99%, Aldrich), ZrO2 (99.0%, Aldrich) and 57Fe enriched Fe2O3. The latter was used to obtain well-resolved 57Fe Mössbauer spectra (see below). Li2CO3 was mixed with the various oxides in the necessary proportions and they were ground intimately together. This mixture was calcinated at 900 °C, reground, pressed into pellets, and sintered at 1050 °C for 16 h and then removed from the furnace.

X-ray powder diffraction (XRPD)

A small fragment taken from the middle of a sintered polycrystalline pellet was ground and used for X-ray powder diffraction. XRPD patterns were collected using a Siemens D8 diffractometer with CuKα radiation. This was done to characterize the synthetic products, to identify all phases present, and to determine the symmetry and the unit-cell dimension of the garnet. Data were collected between 10° and 70° 2θ. The lattice parameter, a0, and grain size were determined by Rietveld refinement using the program Topas V2.1 (Bruker AXS).

Electron probe micro analysis (EPMA)

Chemical analyses were done using a JEOL 8100 SUPERPROBE at the University of Innsbruck. Small polycrystalline chips, taken from the larger pellets, were embedded in an epoxy holder and the surface was ground and then polished using diamond paste. For the analysis, special attention was paid to grain sizes, grain boundaries and textures. Analytical measuring conditions were 15 kV accelerating voltage and a 10 nA beam current. The standards used were synthetic almandine, Fe3Al2Si3O12, for Fe, synthetic La-phosphate, LaP5O14, for La, and synthetic zircon, ZrSiO4, for Zr. Measuring times were 20 s for the peak maximum and 10 s for the background. Wavelength dispersive spectroscopy (WDS) measurements were made to characterize the synthetic products in terms of their compositional homogeneity for the elements La, Zr, and Fe. Twenty point measurements, distributed over 5 different grains, were made and combined to obtain the average composition for a given synthesis product.

Inductively couple plasma optical emission spectroscopy (ICP-OES)

Measurements were made with a Horiba Jobin Yvon Ultima 2 device at the University of Ulm to determine the Li contents. About 50 mg, taken from a sintered pellet, were divided into 2 batches in order to test the analytical reproducibility. The samples were prepared by dissolving 15 mg of garnet in 2 ml of aqua regia (HCl+HNO3) followed by heating to allow complete dissolution of the garnet. Deionized water was then added to obtain 20 ml of solution. Finally, 1 ml aliquots were taken for the ICP-OES measurements.

57Fe Mössbauer spectroscopy

Spectra were recorded using a multichannel analyzer (1024 channels) operating in conjunction with an electromechanical drive system with symmetric triangular velocity shape. The two simultaneously obtained spectra (512 channels each) were folded and evaluated assuming Lorentzian line shapes. Quadrupole splitting distributions, QSDs, were determined using the program Recoil [12]. During the measurement, the source (57Co/Rh, 50 mCi) and absorber were kept at room temperature. Isomer shift values are reported relative to α-iron at RT.

Results

Phase identification and lattice constant (XRPD)

The XRPD patterns of the different synthetic products are shown in Fig. 2, together with the patterns of cubic and tetragonal LLZO [13,14]. The results of the phase identification, the amount of phases and the unit-cell parameters are reported in Table 1 and illustrated in Fig. 3. The diffraction pattern for the syntheses of Li7−3FeLa3Zr2O12 with xint<0.16 exhibit broad and split reflections indicating a mixture of two garnet phases with tetragonal and cubic symmetry. The amount of the cubic phase is about 54% for the composition with xint=0.04 and 74% for the composition with xint=0.08.

There are no indications for the tetragonal garnet phase or of any other phase in synthetic products for compositions with 0.08FeLaO3 (about 0.5 wt%).

In the synthesis product with xint=0.64 about 5% Li2ZrO3 and 10% FeLaO3 were identified. For xint=0.72, 8% Li2ZrO3, 15% FeLaO3, and 1% La2Zr2O7 were determined. The presence of a non-garnet Fe-containing phase for the synthesis with xint=0.52 could be interpreted as indicating that the maximum solubility limit for Fe3+ in LLZO would be about 0.52 Fe pfu, but this is not the case (see below). The a0 value of all cubic garnets was determined to be about 12.980(5) Å. The reflection line widths of Fe-bearing LLZO are broader than those exhibited by Al- and Ga-bearing LLZO [2,15] and they increase, especially at small 2Θ values, with increasing Fe concentration.

Phase identification (EPMA)

Back scattered electron (BSE) photos of polycrystalline chips made with the electron microprobe are shown in Fig. 4. Examination reveals only garnet and no other phases for syntheses with compositions between xint=0.04 and 0.44. The garnet grains for all syntheses are small with diameters of ~10 μm. The porosity of the samples is approximately similar from the Fe-poor compositions up to xint=0.36. When a non-garnet phase is present, the sample density decreases distinctly. The non-garnet phases are FeLaO3 in samples with Fe mole fractions larger than xint=0.44 and La2Zr2O7 as well as Li2ZrO3 in samples with a Fe content higher than xint=0.64.

Crystal chemical formulae (EPMA and ICP-OES)

The chemical analyses and the calculated mole fractions for all Li7−3FeLa3Zr2O12 garnets with intended mole fraction of Fe xint=0.04–0.72, obtained using electron microprobe and ICP-OES measurements, are given in Table 2 and are shown in Fig. 5.

Table 2

EMPA (Fe, La, Zr) and ICP-OES (Li) results for Li7−3FeLa3Zr2O12 garnets with xint=0.04–0.72 in wt% of the oxide.

xint	Fe2O3	Li2Oa	La2O3	ZrO2	Totalb
0.04	0.29(5)	12.13	59.8(5)	30.5(2)	102.7(8)
0.08	0.66(3)	11.66	60.0(5)	30.4(3)	102.7(8)
0.16	1.32(9)	11.49	59.0(3)	29.9(2)	101.6(6)
0.24	1.79(3)	10.72	57.8(4)	29.5(3)	99.8(6)
0.36	1.92(1)	–	57.6(4)	29.2(2)	88.7(8)
0.44	1.89(1)	–	57.4(3)	29.1(2)	88.4(7)
0.52	2.30(7)	–	57.7(1)	29.6(6)	89.5(8)
0.64	2.65(8)	–	57.6(3)	30.3(3)	90.6(7)
0.72	2.48(7)	–	57.6(3)	29.6(6)	89.7(9)

Value measured by ICP-OES. Data were normalized to 100 wt% of the oxide. Values in brackets are the standard deviations based on 20 point analyses.

Total oxide concentration measured by WDS (Fe2O3, La2O3, ZrO2)+ICP-OES (Li2O).

In samples with xint>0.24, the measured Fe values (xobs) in LLZO are distinctly lower than the intended stoichiometric amount, xint, of Fe. The upper solid-solution limit for Fe in LLZO therefore appears to be about xobs=0.25. The measured Li content (Liobs) of garnet agrees well with the Li content of the starting material (Liint). The measured La (Laobs) and Zr (Zrobs) concentrations are close to their intended values in garnet with a maximum variation of ~1% pfu for samples with xint≤0.24. For samples with xint>0.24, Laobs and Zrobs differ more from the intended values. The crystal-chemical formulas for various Li7−3FeLa3Zr2O12 garnets with xint=0.04–0.24, assuming that there are 12 oxygens in the formula unit, are

Formulas for Li7–3FeLa3Zr2O12 solid solutions with xint>0.24 are not given, because the Li content of these samples was not measurable due to the presence of other phases in the synthetic products. Standard deviations for the cations in the crystal chemical formula are reported in Table 3.

Site distribution of Fe (57Fe Mössbauer spectroscopy).

57Fe Mössbauer spectra of four samples taken at RT are displayed in Fig. 6 (middle).

The spectrum of the garnet with xint=0.16 (Fig. 6 – left) exhibits only quadrupole split doublets in the center of the spectrum. An additional 6-line resonant absorption pattern appears in the spectra of the samples with higher Fe contents, indicating an additional phase with a magnetic hyperfine interaction (see Fig. 6 – middle and right).

Spectrum showing a pure quadrupole interaction: We used the spectrum of sample xint=0.16 for making an analysis of a pure quadrupole interaction. After several attempts to fit the spectrum, it became clear that it was not possible to obtain a satisfactory fit using quadrupole split doublets having simple Lorentzian line shapes. Therefore, two quadrupole splitting distributions assuming Voigt line shapes were used [12]. On the basis of their different isomer shift, δ, and quadrupole splitting, ΔEQ, values, they are assigned to Fe3+ located at two different structural sites in LLZO. The most intense Fe3+ doublet with an area, A, of 79% has an isomer shift δ=0.20 mm/s and a quadrupole splitting of ΔEQ=0.96 mm/s. These values are similar to the isomer shift and quadrupole splitting values of Fe3+ at the 24d tetrahedral position in natural silicate and certain synthetic garnets (i.e., Y3(Fe,Ga)5O12), namely, δ=0.20 mm/s and ΔEQ=0.98 mm/s [16,17]. Therefore, this doublet in the spectrum of Fe-bearing LLZO is assigned to Fe3+ at the 24d site. The second most intense Fe3+ quadrupole doublet with an area of 21% of the total resonant absorption results from a super-position of two components with an identical isomer shift value of δ=0.18 mm/s, but with different quadrupole splitting values of ΔEQ=2.30 mm/s and 3.95 mm/s. These two components have absorption areas of 34% and 66% of the total area of the second doublet.

Together the two components have a combined quadrupole splitting value of ΔEQ=2.87 mm/s. The isomer shift of δ=0.18 mm/s indicates tetrahedrally coordinated Fe3+. The large quadrupole splitting values indicate strongly distorted oxygen coordination polyhedra. The question is how this can be interpreted in a crystal chemical sense. In the case of Al-bearing LLZO, the Al3+ cations are distributed over a range of slightly different 96h structural sites that can be displaced towards neighboring empty 24d and 48g sites depending upon the local site occupancies [18]. Thus, depending on the degree of displacement, slightly different local coordination environments around Al3+ arise and they are reflected by the asymmetric and broad 27Al NMR resonance. The situation with Fe3+ in LLZO could be similar.

Therefore, this second complex Mössbauer doublet is assigned to Fe3+ occurring at slightly different 96h positions and having small variations in the local oxygen coordination geometry. This results in a distribution of slightly different electric field gradients and quadrupole splitting distributions. The fit to the spectrum gives a reasonable χ2 value of 1.56.

Spectrum with a magnetic subpattern: The spectra of the samples with xint=0.24, 0.36 and 0.44 exhibit a paramagnetic and a magnetic subpattern. In order to identify the cause of the magnetic subpattern, we evaluated carefully the spectrum of the sample with xint=0.44, which has the most intense sextet contribution of all the synthetic samples (see Fig. 6 – right). The paramagnetic subpattern was evaluated using similar hyperfine parameters as those adopted for the sample with xint=0.16. They are as follows: δ1=0.20 mm/s, ΔEQ1=1.0 mm/s, and A1=60% and δ2=0.18 mm/s, ΔEQ2=3.0 mm/s, and A2=6%. These parameters were kept constant during the fitting procedure of the magnetic subpattern. This approach affords the best approximation of the situation, we think, even though χ2 is rather large with a value of 13.7. Following this, the magnetic subpattern gives an isomer shift of δ=0.36 mm/s, a quadrupole splitting of ΔEQ=0.04 mm/s, a magnetic hyperfine field of HHF=484 kOe, and an area of A3=34% for the total resonant absorption. These parameters agree well with those of nano-crystalline FeLaO3 reported in the literature, and therefore, we assigned this magnetic hyperfine pattern to the phase FeLaO3 [19].

Discussion

This study concentrates on the crystal chemistry of a series of Li7–3Fe3+La3Zr2O12 garnets, the solid-solution behavior of Fe3+ and its site-partitioning behavior. We previously reported that garnet with the composition Li6.47Fe3+0.19La3Zr2O12 can be synthesized with cubic symmetry space group Ia-3d (a0=12.986(4) Å) [11]. Its 57Fe Mössbauer spectrum, evaluated solely with Lorentzian lines, showed that about 96% of the total iron occurred as Fe3+ and 4% as Fe2+. Roughly two-thirds of the Fe3+ cations were assigned to the 24d tetrahedrally coordinated site and about one-quarter to a highly distorted 4-fold coordinated site, most probably 96h, in the LLZO structure. Smaller amounts of Fe3+ and Fe2+, below 5% each, were assigned to other crystallographic sites. Due to their low amounts, a definite assignment was somewhat uncertain.

In this work we synthesized Li7−3Fe3+La3Zr2O12 garnets from xobs=0.03 to xobs=0.25 Fe3+ pfu. We were able to synthesize pure cubic LLZO with xobs=0.12(1) Fe3+ pfu and 6.54 Li pfu, whereby the former is the minimum amount of bulk Fe3+ needed to obtain the pure cubic phase. This Li content agrees with the reported Li content of 6.5 pfu that is necessary to stabilize the cubic modification of Ta- and Nb-doped LLZO [20].

The lattice constant a0=12.980(5) Å in the various garnets does not vary as a function of Fe3+ content and is similar to that reported in our previous study [11]. Less Fe3+ is incorporated in LLZO compared to Al3+ (0.40 pfu [21]) and Ga3+ (about 0.70 pfu [15]) under similar conditions of synthesis. This is possibly caused by increased Coulomb repulsion related to the higher degree of occupation of Fe3+ at 24d compared to the situation with Al3+ and Ga3+ as well as differences in ionic radii.

The XRPD results, alone, would seem to indicate an upper Fe3+ incorporation limit in LLZO of about xint=0.52. However, Fe3+ values measured directly by electron microprobe analysis are equal or less than about xobs=0.25. It is also notable that phases other than LLZO were not observed in BSE images of samples with Fe contents up to xint=0.44, although we determined that the measured Fe contents (xobs) were less than xint.

It appears that 57Fe Mössbauer spectroscopy is, here, better than XRPD for detecting small amounts of non-garnet Fe-containing phases. The Mössbauer spectra of LLZO samples with Fe contents of xint≥0.24 show that some possibly contain poorly or nano-crystalline FeLaO3. This is possible, because crystalline FeLaO3 was detected by both XRPD and WDS analysis in syntheses with larger Fe contents where xint>0.44. It can be argued that small amounts of poorly or nano-crystalline FeLaO3 (and possibly other Fe-free nano-crystalline Zr-bearing phases) could occur at grain boundaries of LLZO, where they are not easily detectable by microprobe analysis or XRPD. We did not measure any Fe2+ in the garnets studied here, as reported in our first study [11]. It is possible that a small amount of FeLaO3 contributed to the spectrum instead of Fe2+ in LLZO.

Conclusion

It is shown by chemical point analyses made on tiny synthetic LLZO crystals, using the electron microprobe, that up to about 0.25 pfu of Fe3+ can be incorporated in the garnet structure. 57Fe Mössbauer measurements show that about 80% of the Fe3+ is located at the 24d site and about 20% at the general 96h site. Below about about xint = 0.16, we were not able to obtain phase pure cubic garnet, which is probably related to compositional inhomogeneity among different grains. In syntheses with intended Fe3+ amounts greater than 0.25 pfu non-garnet Fe3+-bearing phases appear. It is important to note that XRPD and SEM analysis did not show these non-garnet phases for compositions with less than 0.44 Fe pfu. Standard characterization methods are apparently insufficient for fully documenting the precise nature of certain synthetic LLZO samples. This could explain variations in ionic conductivity behavior for LLZO samples with apparently similar chemical compositions and morphologies.

Table 1

Unit-cell parameter of cubic Li7−3FeLa3Zr2O12 with xint=0.04–0.72 and the amount of phases obtained in the synthesis experiments as determined by Rietveld refinement.

xint	a0 [Å]	Cub. LLZOa [%]	Tetr. LLZOa [%]	Li2ZrO3a [%]	FeLaO3a [%]	La2Zr2O7a [%]
0.04	12.980(1)	54	46	0	0	0
0.08	12.980(2)	74	26	0	0	0
0.16	12.977(1)	100	0	0	0	0
0.24	12.984(7)	100	0	0	0	0
0.36	12.981(5)	100	0	0	0	0
0.44	12.980(5)	100	0	0	0	0
0.52	12.981(6)	~99.5	0	0	~0.5	0
0.64	12.979(10)	85	0	5	10	0
0.72	12.981(10)	~77	0	~8	~14	~1

Errors in the phase amounts are small and are not given. There are no indications for the tetragonal garnet phase or of any other phase in the synthetic products for compositions with 0.08

Table 3

Crystal chemical formulas of Li7−FeLa3Zr2O12 garnets with xint=0.04–0.72 in pfua

xint	xobs	Liobs	Laobs	Zrobs
0.04	0.03(2)	6.89	3.05(3)	2.01(4)
0.08	0.06(1)	6.66	3.06(4)	2.01(4)
0.16	0.12(2)	6.54	3.01(2)	1.98(3)
0.24	0.19(5)	6.19	3.02(3)	2.04(6)
0.36	0.19(2)	–	2.94(3)	1.95(6)
0.44	0.17(2)	–	2.87(4)	1.92(3)
0.52	0.21(2)	–	2.84(4)	1.96(3)
0.64	0.25(2)	–	2.97(4)	2.01(4)
0.72	0.23(2)	–	2.94(5)	1.96(3)

Values are calculated using the oxide weight percent values given in Table 2 on the basis of 12 oxygen pfu.